TalkChess.com

Komodo 13.1 doesn't seem to migrate significantly to Lc0 branch, regular version or no-LMR version. The numbers in the matrix are probably more interesting than the dendrogram, but I will post both. I do not understand why is that, it could not be that "Komodo 13.1 is too weak" to show that migratio...

I took the Sim tester which has some 8,000+ quiet positions with unclear best move and tested engines, including versions of Lc0 nets, all at 1 second / position and additionally SF_dev at much longer 10s/position and 30s/poisition. Also, for some check, one Lc0 at 0.2s/position. Hardware: 4 i7 cor...

I took a look at the SIMEX EPDs, and classified them by material on board. There’s 8238 different EPDs, with 16 possible piece conformations, and the count of each type. It’s not a balanced suite across chess positions, which you would probably need to be making conclusive observations other than s...

I took the Sim tester which has some 8,000+ quiet positions with unclear best move and tested engines, including versions of Lc0 nets, all at 1 second / position and additionally SF_dev at much longer 10s/position and 30s/poisition. Also, for some check, one Lc0 at 0.2s/position. Can you zip simila...

I took the Sim tester which has some 8,000+ quiet positions with unclear best move and tested engines, including versions of Lc0 nets, all at 1 second / position and additionally SF_dev at much longer 10s/position and 30s/poisition. Also, for some check, one Lc0 at 0.2s/position. Hardware: 4 i7 core...

Speed ranging from 17 to middle of 30s into 50 kilo nodes per second. Network is 128x10-2019_0812_1326_54_239 - is it good and is there any better at this point? Wrong speeds for that smallnet and 2080 Super. On your GPU, JHorthos 256x20b latest nets are the strongest to rapid TC, and his latest bi...

Ovyron wrote: ↑

Fri Aug 16, 2019 11:12 am

The Rybka Cluster all over again...

Yes, but Lc0 hardly scales beyond 2 GPUs.

I think that it may be interesting if somebody throw away all opening theory and replace it with stockfish analysis with a lot of computer time and see if other books can beat this book. Yes, I maybe will do such a thing in 2 LTC games. Lc0 with "bignet", which seems to agree by far the most with t...

Here are the results on these two suites at exactly 10s/position, Openings200revised (200 positions) and Midgames239 (239 positions) tested today. RTX 2070 GPU, 4 i7 3.8GHz cores. Lc0 JH 320x24b-410 bignet score=163/200 [averages on correct positions: depth=4.6 time=1.04 nodes=8212] score=192/239 [a...

Well, for the first position on Dann's list, it seems that Cfish-dev and Lc0 (42850) both prefer Rac1. I'm curious to try an Lc0 "bignet". What is that and where can I obtain one? Here's Lc0 after 250M nodes: info depth 30 seldepth 82 time 4483899 nodes 131704920 score cp 87 hashfull 620 nps 29372 ...